An improved multi-keyword ranked search over encrypted cloud data

Documents encrypted on the cloud servers can improve security, it also brings about a challenging question for secure query. We use the vector space model (VSM) to obtain the similarity of the documents. In this paper, we use a secure method to support multi-keyword parallel query, which the improved multi-tree [1] can storage the vector value and query similar to the linear. The advantage of using multiple trees is to improve the efficiency of dynamic update (insertion, deletion, inquire) and support parallel query and multi keyword query. In order to further improve the efficiency of the query, we use the naive Bayesian to classify the documents. Based on the user's query keywords have relevance, we can make similar documents stored in the multi-tree more intensive with naive Bayesian classification. In order to strengthen the privacy query, the article [2] proposed two kinds of security index model, one is known cipher model, the other is known background model. In known background model, we randomly add some vocabulary to the hash table to resist statistical attack. By applying our improved index tree, we can achieve near-linear query and dynamic operation of the documents, which improve the operability and take into account the efficiency and accuracy of the query. The security analysis and experiments verify the security and efficiency of the proposed scheme.